Foundation of Reinforcement Learning

Catalog description

This course focuses on theoretical and algorithmic foundations of bandits and reinforcement learning, involving topics including upper confidence bound methods, Thompson sampling, linear and deep contextual bandits, Markov decision process, Q-learning, policy gradient methods, etc. The course targets graduate-level students with a solid mathematical background (linear algebra, probability and statistics, and basic calculus), and a strong research interest in bandits and reinforcement learning. Prerequisite(s): linear algebra, probability and statistics, and basic calculus, or consent of the instructor and director of graduate studies. Credits: 3